import pandas as pd
import numpy as np
from datetime import datetime
pd.set_option('display.max_columns', None)

import warnings
warnings.simplefilter(action='ignore', category=FutureWarning)

from numpy import mean
from numpy import std
from sklearn.datasets import make_regression
from sklearn.model_selection import cross_val_score
from sklearn.model_selection import RepeatedKFold
from sklearn.linear_model import LinearRegression
from sklearn.neighbors import KNeighborsRegressor
from sklearn.tree import DecisionTreeRegressor
from sklearn.svm import SVR
from sklearn.ensemble import StackingRegressor
from matplotlib import pyplot
from sklearn.linear_model import Ridge
from sklearn.model_selection import GridSearchCV
from sklearn.linear_model import Lasso
from sklearn.linear_model import LinearRegression
from sklearn.ensemble import RandomForestRegressor
from sklearn.ensemble import GradientBoostingRegressor
from yellowbrick.regressor import ResidualsPlot
from yellowbrick.regressor import PredictionError
from yellowbrick.regressor import CooksDistance
from sklearn.neural_network import MLPRegressor
import xgboost

ls official_train_test_split

60-40/  70-30/  80-20/

df = pd.read_csv("official_train_test_split/80-20/train_set.csv")
df_test  = pd.read_csv("official_train_test_split/80-20/test_set.csv")

df_test.tail(2)

Clean n transform data¶

df.record_date = pd.to_datetime(df.record_date, format='%Y-%m-%d')
df_test.record_date = pd.to_datetime(df_test.record_date, format='%Y-%m-%d')

icon_onehot_col = pd.get_dummies(df.weather_situation)
df = pd.concat([df, icon_onehot_col], axis=1).drop(["weather_situation"],axis=1)

icon_onehot_col = pd.get_dummies(df_test.weather_situation)
df_test = pd.concat([df_test, icon_onehot_col], axis=1).drop(["weather_situation"],axis=1)

boruta_1 = pd.read_excel("dataset/Boruta result.xlsx", sheet_name="Boruta result 1 ")
boruta_2 = pd.read_excel("dataset/Boruta result.xlsx", sheet_name="Boruta result 2")

vars_1 = boruta_1[boruta_1["Decision "] == "Confirmed"]["Variable"].values
vars_2 = boruta_2[boruta_2["Decision "] == "Confirmed"]["Variable"].values

df.head(2)

Variable 1¶

var1_new = ['Days since first case', 'clear-night',
       'partly-cloudy-night', 'temperature', 'humidity',
       'windSpeed', 'Number of Tweet', 'Sentiments', 'c1_0_action',
       'c1_1_action', 'c1_2_action', 'c1_3_action', 'c2_0_action',
       'c2_1_action', 'c2_2_action', 'c2_3_action', 'c3_0_action',
       'c3_1_action', 'c4_0_action', 'c4_1_action', 'c4_2_action',
       'c4_3_action', 'c4_4_action', 'c5_0_action', 'c5_1_action',
       'c5_2_action', 'c6_0_action', 'c6_1_action', 'c6_2_action',
       'c6_3_action', 'c7_0_action', 'c7_1_action', 'c7_2_action',
       'c8_0_action', 'c8_1_action', 'c8_2_action', 'c8_3_action',
       'c8_4_action', 'e1_0_action', 'e1_1_action', 'e1_2_action',
       'e2_0_action', 'e2_1_action', 'e2_2_action', 'h1_0_action',
       'h1_1_action', 'h1_2_action', 'h2_0_action', 'h2_1_action',
       'h2_2_action', 'h2_3_action', 'h3_0_action', 'h3_1_action',
       'h3_2_action',  'population_density',
       'urbanPopulation', 'healthExpenditure', 'gdp_per_capita',
       'cvd_death_rate', 'diabetes_prevalence', 'handwashing_facilities',
       'hospital_beds_per_thousand', 'female_smokers', 'male_smokers',
       'life_expectancy', 'Outputvalue_lag1','Outputvalue_lag2',
        'Outputvalue_lag3','Outputvalue_lag4','Outputvalue_lag5',
        'Outputvalue_lag6','Outputvalue_lag7',
        'Month','Day_of_Month','Day_of_Week','Log_new_cases_per_million']

df_train = df[var1_new]
df_test = df_test[var1_new]

print(df_train.shape, df_test.shape)

(9207, 76) (2302, 76)

df_train.head(2)

y_train = df_train['Log_new_cases_per_million']
X_train = df_train.drop('Log_new_cases_per_million', axis=1)

y_test = df_test['Log_new_cases_per_million']
X_test = df_test.drop('Log_new_cases_per_million', axis=1)

Train Model¶

def get_stacking():
    # define the base models
    level0 = list()
#     level0.append(('cart', DecisionTreeRegressor()))
    level0.append(('ranFor',RandomForestRegressor(random_state=0)))
    level0.append(('graBoost',GradientBoostingRegressor(random_state=0)))
    level0.append(('ridge', Ridge(alpha = 10,tol=0.001)))
    level0.append(('lasso', Lasso(alpha = 0.001)))
    level0.append(('linear_reg', LinearRegression()))
    # define meta learner model
    level1 = LinearRegression()
    # define the stacking ensemble
    model = StackingRegressor(estimators=level0, final_estimator=level1, cv=5)
    return model

# get a list of models to evaluate
def get_models():
    models = dict()
#     models['cart'] = DecisionTreeRegressor()
    models['ranFor'] = RandomForestRegressor(random_state=0)
    models['graBoost'] = GradientBoostingRegressor(random_state=0)
    models['ridge'] = Ridge(alpha = 0.001)
    models['lasso'] = Lasso(alpha = 0.001)
    models['linear_reg'] = LinearRegression()
    models['stacking'] = get_stacking()
    return models

R2¶

# evaluate a given model using cross-validation
def evaluate_model(model):
    cv = RepeatedKFold(n_splits=10, n_repeats=5, random_state=1)
    scores = cross_val_score(model, X_train, y_train, scoring='r2', cv=cv, n_jobs=-1, error_score='raise')
    return scores

# get the models to evaluate
models = get_models()

# evaluate the models and store results
results, names = list(), list()
for name, model in models.items():
    scores = evaluate_model(model)
    results.append(scores)
    names.append(name)
    print('>%s %.3f (%.3f)' % (name, mean(scores), std(scores)))

# plot model performance for comparison
pyplot.boxplot(results, labels=names, showmeans=True)
pyplot.show()
%matplotlib inline

>ranFor 0.779 (0.016)
>graBoost 0.777 (0.016)
>ridge 0.763 (0.017)
>lasso 0.763 (0.017)
>linear_reg 0.763 (0.017)

/home/harry/anaconda3/lib/python3.7/site-packages/joblib/externals/loky/process_executor.py:706: UserWarning: A worker stopped while some jobs were given to the executor. This can be caused by a too short worker timeout or by a memory leak.
  "timeout or by a memory leak.", UserWarning

>stacking 0.783 (0.016)

Root mean square error¶

# evaluate a given model using cross-validation
def evaluate_model(model):
    cv = RepeatedKFold(n_splits=10, n_repeats=3, random_state=1)
    scores = cross_val_score(model, X_train, y_train, scoring='neg_root_mean_squared_error', cv=cv, n_jobs=-1, error_score='raise')
    return scores

# get the models to evaluate
models = get_models()

# evaluate the models and store results
results, names = list(), list()
for name, model in models.items():
    scores = evaluate_model(model)
    results.append(scores)
    names.append(name)
    print('>%s %.3f (%.3f)' % (name, mean(scores), std(scores)))

# plot model performance for comparison
pyplot.boxplot(results, labels=names, showmeans=True)
pyplot.show()
%matplotlib inline

>ranFor -0.413 (0.016)
>graBoost -0.415 (0.017)
>ridge -0.428 (0.017)
>lasso -0.427 (0.018)
>linear_reg -0.428 (0.017)
>stacking -0.409 (0.016)

Feature Importance¶

reg = RandomForestRegressor(random_state=0)
reg.fit(X_train, y_train)
(pd.Series(reg.feature_importances_, index=X_train.columns)
   .nlargest(10)
   .plot(kind='barh'))

<matplotlib.axes._subplots.AxesSubplot at 0x7fb8adc85310>

Residuals + R2 plot¶

Residuals, in the context of regression models, are the difference between the observed value of the target variable (y) and the predicted value (ŷ), i.e. the error of the prediction. The residuals plot shows the difference between residuals on the vertical axis and the dependent variable on the horizontal axis, allowing you to detect regions within the target that may be susceptible to more or less error.

model = RandomForestRegressor(random_state=0)
visualizer = ResidualsPlot(model)
visualizer.fit(X_train, y_train)
visualizer.show()

<matplotlib.axes._subplots.AxesSubplot at 0x7fb8b137cf10>

from sklearn.inspection import plot_partial_dependence
reg = RandomForestRegressor(random_state=0).fit(X_train, y_train)
disp1 = plot_partial_dependence(reg, X_train,features=[1,2,3])

disp1 = plot_partial_dependence(reg, X_train,features=[3,4,5])

disp1 = plot_partial_dependence(reg, X_train,features=[6,7,8])

disp1 = plot_partial_dependence(reg, X_train,features=[9,10,11])

disp1 = plot_partial_dependence(reg, X_train,features=[12,13,14])

disp1 = plot_partial_dependence(reg, X_train,features=[15,16,17,18])

R2¶

model = RandomForestRegressor(random_state=0).fit(X_train, y_train)
model.score(X_test, y_test)

0.7932450118388492

RMSE¶

from sklearn.metrics import mean_squared_error
y_pred = model.predict(X_test)
mean_squared_error(y_test, y_pred,squared=True)

0.15950357146091934

Variables - 2¶

var2_ = ['Days since first case', 'clear-night',
       'partly-cloudy-day','partly-cloudy-night', 'temperature', 'humidity',
       'windSpeed', 'Number of Tweet', 'Sentiments', 'c1_0_action',
       'c1_1_action', 'c1_2_action', 'c1_3_action', 'c2_0_action',
       'c2_1_action', 'c2_2_action', 'c2_3_action', 'c3_2_action',
       'c4_0_action', 'c4_1_action', 'c4_2_action', 'c4_3_action',
       'c4_4_action', 'c5_0_action', 'c5_1_action', 'c5_2_action',
       'c6_0_action', 'c6_1_action', 'c6_2_action', 'c6_3_action',
       'c7_0_action', 'c7_1_action', 'c7_2_action', 'c8_0_action',
       'c8_1_action', 'c8_2_action', 'c8_3_action', 'c8_4_action',
       'e1_0_action', 'e1_1_action', 'e1_2_action', 'e2_0_action',
       'e2_1_action', 'e2_2_action', 'h1_0_action', 'h1_1_action',
       'h1_2_action', 'h2_0_action', 'h2_1_action', 'h2_2_action',
       'h2_3_action', 'h3_0_action', 'h3_1_action', 'h3_2_action',
       'H5_Investment in vaccines', 'population_density',
       'aged_65_older_sum', 'gdp_per_capita', 'cvd_death_rate',
       'handwashing_facilities', 'hospital_beds_per_thousand',
       'life_expectancy','Log_new_cases_per_million']

df_train = df[var2_]
df_test = df_test[var2_]

print(df_train.shape, df_test.shape)

(9207, 63) (2302, 63)

y_train = df_train['Log_new_cases_per_million']
X_train = df_train.drop('Log_new_cases_per_million', axis=1)

y_test = df_test['Log_new_cases_per_million']
X_test = df_test.drop('Log_new_cases_per_million', axis=1)

Train Models¶

def get_stacking():
    # define the base models
    level0 = list()
    level0.append(('xgboost',xgboost.XGBRegressor()))
    level0.append(('ranFor',RandomForestRegressor(random_state=0)))
    level0.append(('graBoost',GradientBoostingRegressor(random_state=0)))
    level0.append(('ridge', Ridge(alpha = 10,tol=0.001)))
    level0.append(('lasso', Lasso(alpha = 0.001)))
    level0.append(('linear_reg', LinearRegression()))
    # define meta learner model
    level1 = LinearRegression()
    # define the stacking ensemble
    model = StackingRegressor(estimators=level0, final_estimator=level1, cv=5)
    return model

# get a list of models to evaluate
def get_models():
    models = dict()
    models['xgboost'] = xgboost.XGBRegressor()
    models['ranFor'] = RandomForestRegressor(random_state=0)
    models['graBoost'] = GradientBoostingRegressor(random_state=0)
    models['ridge'] = Ridge(alpha = 0.001)
    models['lasso'] = Lasso(alpha = 0.001)
    models['linear_reg'] = LinearRegression()
    models['stacking'] = get_stacking()
    return models

R2¶

# evaluate a given model using cross-validation
def evaluate_model(model):
    cv = RepeatedKFold(n_splits=10, n_repeats=5, random_state=1)
    scores = cross_val_score(model, X_train, y_train, scoring='r2', cv=cv, n_jobs=-1, error_score='raise')
    return scores

# get the models to evaluate
models = get_models()

# evaluate the models and store results
results, names = list(), list()
for name, model in models.items():
    scores = evaluate_model(model)
    results.append(scores)
    names.append(name)
    print('>%s %.3f (%.3f)' % (name, mean(scores), std(scores)))

# plot model performance for comparison
pyplot.boxplot(results, labels=names, showmeans=True)
pyplot.show()
%matplotlib inline

>xgboost 0.769 (0.017)
>ranFor 0.780 (0.016)
>graBoost 0.696 (0.015)
>ridge 0.328 (0.019)
>lasso 0.328 (0.020)
>linear_reg 0.328 (0.019)

A worker stopped while some jobs were given to the executor. This can be caused by a too short worker timeout or by a memory leak.

>stacking 0.786 (0.015)

Root mean squared error¶

# evaluate a given model using cross-validation
def evaluate_model(model):
    cv = RepeatedKFold(n_splits=10, n_repeats=5, random_state=1)
    scores = cross_val_score(model, X_train, y_train, scoring='neg_root_mean_squared_error', cv=cv, n_jobs=-1, error_score='raise')
    return scores

# get the models to evaluate
models = get_models()

# evaluate the models and store results
results, names = list(), list()
for name, model in models.items():
    scores = evaluate_model(model)
    results.append(scores)
    names.append(name)
    print('>%s %.3f (%.3f)' % (name, mean(scores), std(scores)))

# plot model performance for comparison
pyplot.boxplot(results, labels=names, showmeans=True)
pyplot.show()
%matplotlib inline

>ranFor -0.412 (0.015)
>graBoost -0.485 (0.014)
>ridge -0.721 (0.021)
>lasso -0.721 (0.021)
>linear_reg -0.721 (0.021)
>stacking -0.408 (0.015)

Feature Importance¶

reg = RandomForestRegressor(random_state=0)
reg.fit(X_train, y_train)
(pd.Series(reg.feature_importances_, index=X_train.columns)
   .nlargest(10)
   .plot(kind='barh'))

<matplotlib.axes._subplots.AxesSubplot at 0x7fb8acca2150>

Residual + R2 plot¶

model = RandomForestRegressor(random_state=0)
visualizer = ResidualsPlot(model)
visualizer.fit(X_train, y_train)
visualizer.show()

<matplotlib.axes._subplots.AxesSubplot at 0x7fb8ae20a9d0>

R2¶

from sklearn.metrics import r2_score
model = RandomForestRegressor(random_state=0).fit(X_train, y_train)
y_pred = model.predict(X_test)
r2_score(y_test, y_pred)

0.7955888800139136

# model.score(X_test, y_test) # =R2

RMSE¶

from sklearn.metrics import mean_squared_error
y_pred = model.predict(X_test)
mean_squared_error(y_test, y_pred,squared=True)

0.1576953667434353

from sklearn.inspection import plot_partial_dependence
disp1 = plot_partial_dependence(model, X_train,features=[1,2,3])

disp1 = plot_partial_dependence(reg, X_train,features=[3,4,5])

disp1 = plot_partial_dependence(reg, X_train,features=[6,7,8])

disp1 = plot_partial_dependence(reg, X_train,features=[9,10,11])

disp1 = plot_partial_dependence(reg, X_train,features=[12,13,14])

disp1 = plot_partial_dependence(reg, X_train,features=[15,16,17])

disp1 = plot_partial_dependence(reg, X_train,features=[18,19,20])

disp1 = plot_partial_dependence(reg, X_train,features=[21,22,23])

disp1 = plot_partial_dependence(reg, X_train,features=[24,25,26])

disp1 = plot_partial_dependence(reg, X_train,features=[27,28,29])

disp1 = plot_partial_dependence(reg, X_train,features=[30,31,32])

disp1 = plot_partial_dependence(reg, X_train,features=[33,35,36])

disp1 = plot_partial_dependence(reg, X_train,features=[37,38,39])

disp1 = plot_partial_dependence(reg, X_train,features=[40,41,42])

disp1 = plot_partial_dependence(reg, X_train,features=[43,44,45])

disp1 = plot_partial_dependence(reg, X_train,features=[46,47,48])

disp1 = plot_partial_dependence(reg, X_train,features=[49,50,51])

disp1 = plot_partial_dependence(reg, X_train,features=[52,53,54])

disp1 = plot_partial_dependence(reg, X_train,features=[55,56,57])

disp1 = plot_partial_dependence(reg, X_train,features=[58,59,60])

SHAP (SHapley Additive exPlanations)¶

import shap

# load JS visualization code to notebook
shap.initjs()

# explain the model's predictions using SHAP
# (same syntax works for LightGBM, CatBoost, scikit-learn and spark models)
explainer = shap.TreeExplainer(model)
shap_values = explainer.shap_values(X_train)

# visualize the first prediction's explanation (use matplotlib=True to avoid Javascript)
shap.force_plot(explainer.expected_value, shap_values[0,:], X_train.iloc[0,:])

The above explanation shows features each contributing to push the model output from the base value to the model output. Features pushing the prediction higher are shown in red, those pushing the prediction lower are in blue. The base_value here is 0.4979 while our predicted value is 0.7. Goal Scored = 2 has the biggest impact on increasing the prediction, while ball possession feature has the biggest effect in decreasing the prediction.

# visualize the training set predictions
shap.force_plot(explainer.expected_value, shap_values, X_train)

shap.plots.force is slow for many thousands of rows, try subsampling your data.

# summarize the effects of all the features
shap.summary_plot(shap_values, X_train)

To get an overview of which features are most important for a model we can plot the SHAP values of every feature for every sample. The summary plot tells which features are most important, and also their range of effects over the dataset.

Local Surrogate (LIME)¶

import lime
import lime.lime_tabular

explainer = lime.lime_tabular.LimeTabularExplainer(X_train.values, feature_names=X_train.columns, verbose=True, mode='regression')

X_test.iloc[25]

Days since first case           106.000
clear-night                       0.000
partly-cloudy-day                 1.000
partly-cloudy-night               0.000
temperature                      62.610
                                ...    
gdp_per_capita                46682.515
cvd_death_rate                  114.767
handwashing_facilities           97.164
hospital_beds_per_thousand        2.500
life_expectancy                  80.900
Name: 25, Length: 62, dtype: float64

i = 25
exp = explainer.explain_instance(X_test.iloc[i], model.predict, num_features=5)

Intercept 0.12229108069782985
Prediction_local [0.69854288]
Right: 0.6892780249531296

exp.show_in_notebook(show_table=True)

exp.as_list()

[('gdp_per_capita > 34272.36', 0.24783541975768708),
 ('aged_65_older_sum > 12.82', 0.16215411476367797),
 ('cvd_death_rate <= 137.02', 0.15614843314903928),
 ('c2_0_action <= 0.00', 0.0993510132792419),
 ('c8_4_action <= 0.00', -0.08923718243084912)]

	CountryCode	Days since first case	record_date	c1_0_action	c1_1_action	c1_2_action	c1_3_action	c2_0_action	c2_1_action	c2_2_action	c2_3_action	c3_0_action	c3_1_action	c3_2_action	c4_0_action	c4_1_action	c4_2_action	c4_3_action	c4_4_action	c5_0_action	c5_1_action	c5_2_action	c6_0_action	c6_1_action	c6_2_action	c6_3_action	c7_0_action	c7_1_action	c7_2_action	c8_0_action	c8_1_action	c8_2_action	c8_3_action	c8_4_action	e1_0_action	e1_1_action	e1_2_action	e2_0_action	e2_1_action	e2_2_action	h1_0_action	h1_1_action	h1_2_action	h2_0_action	h2_1_action	h2_2_action	h2_3_action	h3_0_action	h3_1_action	h3_2_action	E3_Fiscal measures	E4_International support	H4_Emergency investment in healthcare	H5_Investment in vaccines	Log_new_cases_per_million	Outputvalue_lag1	Outputvalue_lag2	Outputvalue_lag3	Outputvalue_lag4	Outputvalue_lag5	Outputvalue_lag6	Outputvalue_lag7	population_density	gdp_per_capita	cvd_death_rate	diabetes_prevalence	handwashing_facilities	hospital_beds_per_thousand	female_smokers	male_smokers	life_expectancy	aged_65_older_sum	humidity	weather_situation	temperature	windSpeed	Number of Tweet	Sentiments	isHoliday	urbanPopulation	healthExpenditure	Month	Day_of_Month	Day_of_Week
2300	NOR	41	2020-04-08	0	0	0	28	0	30	0	0	0	0	28	0	0	0	0	16	0	28	0	42	0	0	0	0	0	24	0	0	0	0	25	42	0	0	0	0	27	0	0	42	0	42	0	0	0	8	0	0.0	0.0	0.0	0.0	1.326602	1.379849	1.74591	1.70206	1.697264	1.604356	1.610298	1.359323	14.462	64800.057	114.316	5.31	86.979	3.6	19.6	20.7	82.40	13.817	0.65	partly-cloudy-day	41.09	7.75	1.0	0.6124	-1	82.616	8.928808	4	8	2
2301	FJI	70	2020-05-29	0	0	0	72	0	0	15	0	0	0	75	0	0	0	72	0	15	0	0	0	0	72	0	0	0	72	0	0	0	0	62	0	64	0	0	0	65	0	0	129	0	0	0	43	0	0	72	0.0	0.0	0.0	0.0	0.000000	0.000000	0.00000	0.00000	0.000000	0.000000	0.000000	0.000000	49.562	8702.975	412.820	14.49	65.386	2.3	10.2	34.8	67.44	4.754	0.78	cloudy	77.03	8.27	0.0	0.0000	-1	56.750	2.318574	5	29	4

	CountryCode	Days since first case	record_date	c1_0_action	c1_1_action	c1_2_action	c1_3_action	c2_0_action	c2_1_action	c2_2_action	c2_3_action	c3_0_action	c3_1_action	c3_2_action	c4_0_action	c4_1_action	c4_2_action	c4_3_action	c4_4_action	c5_0_action	c5_1_action	c5_2_action	c6_0_action	c6_1_action	c6_2_action	c6_3_action	c7_0_action	c7_1_action	c7_2_action	c8_0_action	c8_1_action	c8_2_action	c8_3_action	c8_4_action	e1_0_action	e1_1_action	e1_2_action	e2_0_action	e2_1_action	e2_2_action	h1_0_action	h1_1_action	h1_2_action	h2_0_action	h2_1_action	h2_2_action	h2_3_action	h3_0_action	h3_1_action	h3_2_action	E3_Fiscal measures	E4_International support	H4_Emergency investment in healthcare	H5_Investment in vaccines	Log_new_cases_per_million	Outputvalue_lag1	Outputvalue_lag2	Outputvalue_lag3	Outputvalue_lag4	Outputvalue_lag5	Outputvalue_lag6	Outputvalue_lag7	population_density	gdp_per_capita	cvd_death_rate	diabetes_prevalence	handwashing_facilities	hospital_beds_per_thousand	female_smokers	male_smokers	life_expectancy	aged_65_older_sum	humidity	temperature	windSpeed	Number of Tweet	Sentiments	isHoliday	urbanPopulation	healthExpenditure	Month	Day_of_Month	Day_of_Week	clear-day	clear-night	cloudy	fog	partly-cloudy-day	partly-cloudy-night	rain	snow	wind
0	CAN	112	2020-05-17	0	0	0	63	0	0	0	61	0	0	67	0	0	0	0	56	113	0	0	0	65	0	0	0	0	59	0	0	0	0	61	0	0	64	0	61	0	0	0	68	0	0	0	60	0	113	0	0.0	0.0	0.0	0.0	1.502714	1.47356	1.472786	1.493584	1.477396	1.482359	1.526288	1.602722	4.037	44017.591	105.599	7.37	89.443	2.5	12.0	16.6	82.43	13.8905	0.72	51.15	4.7	1.0	-0.361200	-1	81.482	7.794102	5	17	6	0	0	0	0	0	1	0	0	0
1	NPL	75	2020-04-09	0	0	0	22	0	0	0	19	0	0	23	0	0	0	0	19	0	0	17	0	0	18	0	0	0	19	0	0	0	0	18	0	11	0	0	0	11	0	0	73	0	88	0	0	0	0	77	0.0	0.0	0.0	0.0	0.000000	0.00000	-0.987163	0.000000	-1.468521	0.000000	0.000000	0.000000	204.430	2442.804	260.797	7.26	47.782	0.3	9.5	37.8	70.78	4.5105	0.43	72.63	2.2	3.0	0.084421	-1	20.153	1.240623	4	9	3	1	0	0	0	0	0	0	0	0